Although it is common to assess the language proficiencies of nonnative speaker students using standardized tests in educational contexts where language proficiency tends to be considered a determiner of academic success, some studies have argued that language proficiency should instead be considered a concept of communicative competence (e.g., Bachman, 1990; Canale & Swain, 1980; Celce-Murcia et al., 1995). As these studies have noted, however, most Two-Way Immersion (TWI) research regarding students’ language proficiencies has been based on standardized tests (e.g., Lindholm-Leary, 2001; Oller & Eilers, 2002; Thomas & Collier, 2002). Although various studies have found that students in TWI programs demonstrated progress toward the goals of bilingualism and biliteracy development (e.g., Alanis, 2000; de Jong, 2004), the outcomes in terms of the non-English language development are not consistent (e.g., Ha, 2001; Kanagy, 2001; Kovelman et al., 2008). Therefore, this study attempts to explore an alternative way of assessing students’ language proficiencies beyond using standardized tests in order to better understand their bilingual development.Based upon sociocultural theories of learning, this study adopts the concept of perceived proficiency, i.e., that language proficiency is perceived by participants who constitute learning through interaction. In particular, this study focuses on two bilingual teachers’ perceived proficiencies of students in a Korean/English TWI program. In addition, it works toward assessing students’ bilingual language proficiency by analyzing FLOSEM scores and students’ use of referential choices in order to examine the relationship with teachers’ perceived proficiencies. Last, this study investigates how teachers’ perceived proficiencies influence classroom practices using positioning theory.The data used in this study were collected through video and audio recordings of classroom interactions (24 hours); four interviews (two with each teacher); and students’ narratives in both English and Korean (eight narratives; two narratives per student). For the investigation of teachers’ perceived proficiencies of students, 46 interview excerpts were selected based on the researchers’ questions about students’ language ability. Clancy’s (1995) categorization of lexical forms was used for the analysis of referential choice. The findings revealed that the two teachers’ perceived language proficiencies of students were constructed according to the specific context, a Korean/English TWI classroom. Mainly, both teachers considered the acquisition of ‘content vocabulary’ or ‘content knowledge’ to be the main essential competence for a high level of proficiency in both languages. Along with the perceived proficiencies of students, the teachers’ self-perception about their own Korean proficiencies, particularly their lack of discipline-specific vocabulary, influenced turn-taking processes that may have afforded or constrained students the opportunity to participate during the Korean instructional time. However, a detailed analysis of students’ use of referential choices in both languages and the results of the FLOSEM test suggested student proficiencies that differed from the teachers’ perceived proficiencies. Referential choice involves distinctive language-specific characteristics for each language; yet the teachers did not observe the skillful use of referential choice in the narrative of a student who had not been recognized by the teachers as possessing content vocabulary and knowledge. Throughout the classroom observation video data recorded during the Korean instructional time, teachers’ perceived student language proficiencies played an important role in their designating more proficient students who could model for less proficient students, despite the fact that the more proficient student may not have been consistently proficient according to different ways of measuring proficiency. On the basis of these findings, I discuss implications for theory and practice. Overall, I argue for various approaches to assessing bilingual students’ language proficiencies, particularly in educational contexts such as TWI programs, emphasizing the development of tools that enable teachers to expand the scope of their perceptions about students’ bilingual language proficiency to enhance students’ bilingual language development.